Due to their potential as thermoelectric materials [1,2], intermetallic clathrates (A₈[X,Y]₄₆)—aka type-I; A₂₄[X,Y]₁₃₆)—aka type-II, where A = alkali or alkaline-earth metal; X and Y = groups 13 and 14 elements) have attracted much attention in recent years. Not long ago, our group had reported on several new Si-based clathartes with type-II structure [3]. We were intrigued by the scarcity of isotypic Ge- and particularly Sn-based compounds, with Ba₁₆Ga₃₂Sn₁₀₄ [4] being the sole example. Our attention was also piqued by the fact that the system Ba–Ga–Sn appeared carefully mapped out—for example, the polymorphic Ba₈Ga₁₆Sn₃₀ clathrate-I and clathrate-VIII compounds [5-7], as well as some other ternary phases like BaGaSn [8], Ba₃Ga_0.491Sn_4.509 [9], and BaGa_3.11Sn_0.89 [10], have been well characterized—while in the Ba–In–Sn system, no ternary phases have been obtained until now.

Therefore, we embarked on systematic investigations of the system Ba–In–Sn. The initial results from our exploratory work in the Sn-rich part of this ternary phase diagram suggested that the known binary phases BaSn₃ [11-13] and BaSn₅ [11,14] are the most common products of such reactions. Being at the border of the typical intermetallic and Zintl phases, both compounds are reported as superconductors below 2.4 K (BaSn₃) and 4.4 K (BaSn₅), respectively. The first and only (so far) ternary phase, the solid-solution BaSn_3−xIn_x was identified from a reaction of the elements using excess In as a metal flux. Its structure is the main subject of this article and was refined from single-crystal X-ray diffraction data; the composition was independently obtained via elemental micro-probe analysis by means of energy-dispersive X-ray spectroscopy (EDX). Electrical resistivity and heat-capacity measurements on single crystals of both phases are also reported.

During the investigations of the Ba–In–Sn system, single crystals of the known barium tristannide BaSn₃ [11-13] and the In-doped barium tristannide BaSn_3−xIn_x (x ≈ 0.2) were obtained, instead of the desired, still unknown, clathrates Ba₈In₁₆Sn₃₀ or Ba₁₆In₃₂Sn₁₀₄. While BaSn₃ can be readily synthesized from elemental mixtures [11-13] or by using Sn flux (this work), BaSn_3−xIn_x with a small (less than ca. 10% at.) substitution of Sn by In was only observed when using large amounts of In as a metal flux. BaSn₃ and BaSn_3−xIn_x both crystallize with the hexagonal space group P6₃/mmc (No. 194) (Pearson symbol hP8; Ni₃Sn type structure [15]), the hexagonal variant of the Cu₃Au structure [16]. The structure can be viewed as alternating layers of Ba²⁺ cations (Wyckoff site 2c: ⅓ ⅔ ¼) and [Sn₃]²⁻ triangles (Wyckoff site 6h: 2x, x, ¾), as shown in Figure 1. These slabs are arranged in an ABAB sequence along the crystallographic c-axis. Since the triangular Sn₃ fragments can also be considered as parts of Sn₆ octahedra that are linked via face-sharing, one can also rationalize the structure as polyanionic 1D-chains along the c-axis and Ba²⁺ cations [13]. Full description of the crystal and electronic structure can be found elsewhere [12,13].

The lattice parameters we report for Sn flux-grown BaSn₃ are a = 7.228(2) Å, c = 5.469(3) Å; the literature values for BaSn₃ [12,13] are a = 7.254(4) and c = 5.500(3) Å. These periodicity constants appear to be slightly longer, however, this elongation is not due to any phase width, but to a thermal expansion—our data were obtained at 120(2) K (Table 1), while the previously reported structure was done at room temperature [12,13]. The virtually identical c/a ratios (0.757 vs 0.758) confirm that BaSn₃ is a line compound with a robust structure, which is invariant of the method of preparation (stoichiometric reactions vs Sn flux).

However, upon comparison of the above values with the refined unit cell parameters for the “same” compound, grown from In flux, one immediately notices that the structure expands slightly in the ab-plane (a = 7.260(1) Å) and contracts along the c axis (c = 5.382(2) Å), which results in significantly smaller c/a ratio of 0.741, i.e., solid solution BaSn_3−xIn_x forms. Given the large excess of In employed in the synthesis, we can speculate that the homogeneity range of this phase is rather small, and that the discussed composition represents the maximum solubility range under these conditions, i.e., BaSn_3−xIn_x (0 ≤ x ≤ 0.2).

In BaSn₃, the Sn–Sn distances within the Sn₃ triangles are 3.051(1) Å, slightly shorter than the corresponding distances in BaSn_3−xIn_x (d_{Sn/In−Sn/In} = 3.075(1) Å). The opposite trend is observed for the interlayer Sn–Sn distances—they are slightly longer in the pure binary compound (d_Sn–Sn = 3.253 Å) compared to the In-doped variant (d_{Sn/In−Sn/In} = 3.224(1) Å). The substitution of Sn with In has a very minor effect (or almost no-effect) on the Ba–Sn and Ba–Sn/In distances—d_Ba–Sn = 3.629(1)/3.646(1) Å vs d_Ba–Sn/In = 3.617(1)/3.644(1) Å, respectively. The discussed bond lengths are in good agreement with those reported for other binary compounds like BaSn (d_Sn−Sn = 2. 985 Å; d_Ba−Sn = 3.600−3.705 Å) [17]; BaSn₂ (d_Sn−Sn = 2.919 Å; d_Ba−Sn = 3.473 Å) [18], BaSn₅ (d_Sn−Sn = 2.959−3.437 Å; d_Ba−Sn = 3.549−3.729 Å) [14], and Ba₃Sn₅ (d_Sn−Sn = 2.992−3.695 Å; d_Ba−Sn = 3.488−3.731 Å) [19].

Electrical resistivity measurements (Figure 2) for single-crystals of BaSn₃ show metallic behavior with ρ_300K on the order of 200 μΩ·cm. At temperature below ca. 4.5 K, the material undergoes a phase change, following which it becomes superconducting, as evidenced from the abrupt jump in the ρ(T) curve from nearly 12 μΩ·cm to zero resistivity at 3.7 K. This behavior is somewhat puzzling, since superconductivity was reported originally below 2.4 K by Fässler et al. [12,13]. Perhaps even more puzzling is the fact that the ρ(T) results are reproducible between crystals from different batches, while heat capacity measurements on the same crystals (supporting information) show no anomalies at this temperature regime. A reasonable interpretation of the data can be accomplished if one recognizes that T_C is very close to the critical temperature of elemental Sn (3.7 K [20]), and since we cannot rule out the presence of small amounts of free Sn in the BaSn₃ specimens grown from Sn flux, most likely this material is not a superconductor.

The resistivity of single-crystals of the In-doped BaSn_2.8In_0.2 confirms similar metallic behavior with ρ_300K on the order of 230 μΩ·cm). This material also becomes a superconductor, but at a higher temperature (ca. 6 K). This T_C is 2−3 degrees higher than the critical temperature of elemental Sn (3.7 K [20]) and In (3.4 K [20]), however the heat capacity measurements mirror the behavior observed for BaSn₃ and show no anomalies at this temperature regime. Although the ρ(T) behavior of another BaSn_3−xIn_x sample is similar (T_C is slightly different but the sample might inadvertently has slightly different doping level), the susceptibility data indicate only a small volume fraction. Therefore, we argue that bulk superconductivity is not an intrinsic property of these materials.

Both BaSn₃ and BaSn_2.8In_0.2 were obtained using excess of Sn and In as metal fluxes, respectively. In all cases, the corresponding elements (Alfa or Aldrich, purity greater than 99.9%) were loaded in alumina crucibles (2 cm³) inside an argon-filled glove box. After enclosing the crucibles in evacuated fused silica tubes, they were heated in programmable muffle furnaces to 1000 °C (rate 10 °C/h), equilibrated for 5 h, cooled to 450 °C (rate −300 °C/h) and annealed at 450 °C for 12 h. The flux was removed (by decanting it) subsequent to another cooling step to 400 °C (rate −10 °C/h). The silica tubes were crack-opened in the glove box and the crystals were isolated.

BaSn₃ crystals grow very big (up to 7–8 mm long) from Sn flux reactions (loaded in the molar ratio Ba:Sn = 2:15), whereas similar in size crystals of BaSn_3−xIn_x emerge only from In flux reactions (loaded in the molar ratio Ba:In:Sn = 2:15:7.5). In both cases, the crystals' appear as silver rods. Extensive efforts to synthesize the desired clathrate Ba₈In₁₆Sn₃₀ (or any other _ternary phase in this system) have been unsuccessful so far.

X-ray powder diffraction patterns were taken at room temperature on a Rigaku MiniFlex powder diffractometer using Cu Kα radiation. Typical runs included θ–θ scans (2θ_max = 80°) with scan steps of 0.05° and 5 sec/step counting time. The JADE 6.5 software package [21] was used for data analysis. The intensities and the positions of the experimentally observed peaks and those calculated from the crystal structures matched very well. Since the diffractometer was enclosed in a nitrogen-filled glove box, we were able to test the air-sensitivity of the title compounds by comparing the diffraction patterns of the freshly prepared samples with those that had been exposed to air. Based on the test results, all of the title compounds are moderately unstable and decompose/hydrolyze after an overnight exposure to air.

Intensity data collections were carried out on a Bruker SMART CCD diffractometer at 120 K using graphite-monochromated Mo-Kα radiation. Suitable single-crystals of BaSn₃ and BaSn_3−xIn_x were selected in a glove box and cut to smaller dimensions (less than 0.1 mm) under mineral oil. The SMART [22] and SAINTplus [23] programs were used for the data collection, integration and the global unit cell refinement from all data. Semi-empirical absorption correction was applied with SADABS [24]. The structures were refined to convergence by full matrix least-square methods on F², as implemented in SHELXTL [25]. All atoms were refined with anisotropic displacement parameters.

In the refinements of BaSn_3−xIn_x, the Sn atoms, located at Wyckoff site 6h, were determined to be statistically disordered with a small amount of In (vide supra). This was done by freeing the occupation parameter of the Sn site, while keeping the Ba-one fixed, which suggested a slight deviation from full occupancy. Because the X-ray scattering factors for In and Sn are very close, reliable refinements of the composition were not possible and the ratio of Sn to In was established from elemental microanalysis (JEOL 7400F electron microscope equipped with an INCA-OXFORD energy-dispersive spectrometer). In agreement with the averaged ratio of Sn:In = 2.8:0.2, the 6h site was assigned as 7 atomic % In and the balance Sn (fixed), yielding a formula BaSn_2.8In_0.2. Selected details of the data collections and structure refinement parameters are summarized in Table 1. The refined atomic coordinates and equivalent isotropic displacement parameters are given in Table 2. Additional details of the crystal structure investigations may be obtained from Fachinformationszentrum Karlsruhe, D-76344 Eggenstein-Leopoldshafen, Germany (Fax: +49-7247-808-666, E-Mail: crysdata@fiz-karlsruhe.de) on quoting the depository numbers CSD-423108 for BaSn₃ and CSD-423109 for BaSn_2.8In_0.2, respectively.

Four-probe measurements of the electrical resistivity and heat capacity measurements (thermal relaxation method) as a function of the temperature were carried out on a Quantum Design PPMS system in the interval 2 to 300 K with excitation current of 1 mA. The electrical resistivity was measured on at least two crystals from each batch to assure reproducibility. Polished single crystals were used to minimize geometric errors.

Large single-crystals of BaSn₃ and BaSn_2.8In_0.2 were obtained using the flux method, following systematic investigations of the Ba–In–Sn ternary diagram. The first and only (so far) ternary phase, the solid-solution BaSn_3−xIn_x was identified from a reaction of the elements using excess in as a metal flux, while BaSn₃ crystals can be readily grown from Sn flux. Both structures were refined from single-crystal X-ray diffraction data and the composition of BaSn_2.8In_0.2 was independently confirmed via energy-dispersive X-ray spectroscopy (EDX). Electrical resistivity measurements on single crystals of both phases indicate metallic behavior in the temperature interval 6−300 K and superconducting behavior below 4–6 K. However, the susceptibility data indicate only a small volume fraction of a superconducting phase in the samples. Based on this, and the fact that the heat capacity measurements show no anomalies around 2–10 K, we argue that bulk superconductivity is not an intrinsic property of these materials.